It remembers your business, not your chat.
Persistent memory plus skills the agent writes and improves itself. State survives every session — margins, vendors, the deal memo from March, who owns what. You never re-explain the company again.
GREA Agent is your CXO's chief of staff. It holds the company context permanently, runs the recurring work on a schedule while you sleep, and reaches you on Telegram, Slack or Discord. One flat cost. Your infrastructure. No usage windows.
Persistent memory plus skills the agent writes and improves itself. State survives every session — margins, vendors, the deal memo from March, who owns what. You never re-explain the company again.
An always-on host plus cron jobs: a 07:00 operating brief, KPI watch, AR chase, vendor-spend review. Nobody has to open a tab. The work is simply done and waiting for you.
Plain files on disk — config.yaml, SOUL.md, skills/, .env. Move hosts in minutes. Credentials never leave the owner's machine.
A seat licence is priced per human who touches the tool — so the assistant gets more expensive exactly as it becomes more useful to the whole firm. GREA prices the work, not the headcount.
On the seat model, the same team costs $300 now and $1,500 the moment anyone needs real usage. GREA Firm is flat at $299 regardless of headcount — and it does not reset every five hours.
Every figure in the Claude column was read live from Anthropic's own pricing page. Nothing here is estimated.
| Capability | Claude Team or Enterprise | GREA Agent |
|---|---|---|
| Pricing model | Per seat, per month. Team Standard $20/seat/month billed annually, $25 billed monthly; Premium seat (5× usage) $100/seat/month annual, $125 monthly. Enterprise seat price + usage at API rates (US$20/seat/month billed annually). | Flat monthly price for the whole firm. Operator $99, Firm $299, Enterprise from $1,500. No seat count. |
| Usage ceiling | Limits reset on a rolling five-hour session window; paid plans add weekly limits. Anthropic "may limit your usage in other ways, such as weekly and monthly caps or model and feature usage, at our discretion." | No usage windows. The agent runs on your infrastructure; there is no 5-hour clock to beat. |
| Company context | Conversation-scoped by default. Continuity depends on memory features and the thread you are in. | Persistent memory plus skills the agent writes itself. State survives every session, permanently. |
| Autonomy & scheduling | You open it and drive it. Managed Agents add-ons bill $0.08 per session-hour for active runtime. | Always-on host plus cron: 07:00 brief, KPI watch, AR chase, vendor-spend review — runs while nobody is present. |
| Ownership & portability | Platform-hosted. Team plan has no ACH and no invoicing or net terms. Enterprise adds SCIM, audit logs, role-based access, a HIPAA-ready offering, custom data retention controls, and network-level access control / IP allowlisting. | Plain files on disk — config.yaml, SOUL.md, skills/, .env. Move hosts in minutes. Credentials never leave the owner's machine. |
| Where it reaches you | Anthropic's own web, desktop and mobile apps. | Telegram, Slack or Discord — the channels your team already lives in. |
| Data & training | "No model training on your content by default" on Team and Enterprise — an opt-out, not a structural guarantee. | Content stays on the owner's infrastructure. Nothing to opt out of. |
| The model | Anthropic's frontier models — Opus 5.5 (API $4 / $20 per MTok in/out), Sonnet 5.5 ($2 / $10), Haiku 5.5 ($0.10 / $0.50 for prompts ≤ 100K tokens). | Same-or-comparable routed models. GREA is not a frontier lab and does not claim a better model — see What this is not below. |
Figures read from claude.com/pricing on 2026-10-11
We are not going to tell you otherwise. If the job is the hardest single reasoning problem you have ever posed, the frontier labs sell a better raw model. That is not what this product is.
What we do claim: same-or-comparable model, better agent for a business. It owns the company context permanently, it runs on a schedule without anyone present, it lives on infrastructure the owner controls, and it has no usage ceiling. That is the product. The model is table stakes.
The seat math, concretely: the same work costs $25/seat/month on a Claude Team Standard seat and $125/seat/month on a Premium seat. Twelve people is $300/mo today and $1,500/mo the moment usage matters. GREA Firm is flat at $299 regardless of headcount.
These ship with the product — not slideware. Each one is an operating routine the agent runs on your schedule.
This box calls the real GREA Agent demo API — Workers AI, live, no mock. It answers as a CXO chief of staff would.
Operators and firms first. We onboard in small batches so every agent gets real skills, not a demo.
—
If you already run Hermes, install the profile directly:
$ hermes profile install github.com/grea/grea-agent --name atlas
Name it whatever you want — it becomes your agent's name.